How TypeTez works out your typing speed

Typing tests disagree about what a "word" is and what an error costs, so two sites can give the same person different numbers. Here is exactly what TypeTez counts, with worked examples you can check by hand.

The three numbers in your result

Gross WPM

(characters in every word you typed + one space per word) ÷ 5 ÷ minutes

Net WPM

(characters in the words you typed exactly right + one space per word) ÷ 5 ÷ minutes

Accuracy

net characters ÷ gross characters × 100

A word is anything between spaces, including its punctuation, so today. needs its full stop to count as right. A word with any wrong, missing or extra letter is a wrong word and adds nothing to your net speed. Capital letters matter too.

Why divide by 5?

Words have different lengths, so TypeTez treats every 5 characters as one standard word. The SSC CHSL 2026 notice implies the same ratio: it says that 35 words per minute and 30 words per minute correspond to about 10,500 and 9,000 key depressions per hour. Working backwards: 10,500 ÷ 60 = 175 key depressions a minute, and 175 ÷ 35 = 5 per word. The Hindi figure gives the same answer: 9,000 ÷ 60 ÷ 30 = 5.

Source: Notice: Combined Higher Secondary (10+2) Level Examination, 2026, Staff Selection Commission (SSC), para 13.9.13.7. Checked 26 October 2026.

Worked example in English

Passage: The rain stopped and the children ran outside.
You type: The rain stoped and the children ran outside. and finish in 15 seconds (0.25 minutes).

• Gross characters = 4 + 5 + 7 + 4 + 4 + 9 + 4 + 9 = 46. Gross WPM = 46 ÷ 5 ÷ 0.25 = 36.8.
• Net characters = 46 − 7 = 39. Net WPM = 39 ÷ 5 ÷ 0.25 = 31.2.
• Accuracy = 39 ÷ 46 × 100 = 84.8%.

One misspelt word cost 5.6 WPM here. In a short test every error is expensive, which is why accuracy comes before speed in our practice plan.

Worked example in Hindi

Passage: हम रोज सुबह पार्क जाते हैं।
You type: हम रोज सुबह पार्क जाते है। (the last word is missing its ं anusvara) in 15 seconds.

Characters are Unicode characters after NFC normalisation, so vowel signs and the virama count separately: हम 2, रोज 3, सुबह 4, पार्क 5 (प, ा, र, ्, क), जाते 4, and the typed है। 3.

• Gross characters = (2+1) + (3+1) + (4+1) + (5+1) + (4+1) + (3+1) = 27. Gross WPM = 27 ÷ 5 ÷ 0.25 = 21.6.
• Net characters = 27 − 4 = 23. Net WPM = 23 ÷ 5 ÷ 0.25 = 18.4.
• Accuracy = 23 ÷ 27 × 100 = 85.2%.

Skipped and extra words

If you skip a word, a simple position-by-position check would mark every later word wrong. TypeTez lines up your words with the passage using the fewest possible changes (an edit distance over whole words), the way a person marking a paper would. Each passage word you typed is then right, wrong or missed, and any word not in the passage is extra.

Example: passage please send the file today, typed please the file today. TypeTez reports 4 correct words and 1 missed word (send). A position-by-position check would have called the, file and today wrong as well.

Missed words do not lower gross or net speed, because nothing was typed for them, but they are listed in your result and counted in the mistake tally.

Unicode normalisation

The same letter can be stored in more than one way. The Devanagari letter क़ can be one character (U+0958) or क followed by the nukta sign (U+0915 U+093C). Unicode Normalization Form C (NFC) gives canonically equivalent text one standard form; for U+0958 the Unicode standard lists it as a composition exclusion, so NFC always uses the two-character form. TypeTez normalises both the passage and your typing to NFC before comparing, so your keyboard's choice never counts against you.

Source: Unicode Standard Annex #15: Unicode Normalization Forms, Unicode Consortium. Checked 26 October 2026.

We also make three small adjustments before comparing: curly quotation marks become straight ones (phones often insert curly quotes), the invisible joiner characters ZWJ and ZWNJ are removed because they only change how a conjunct is drawn, and any run of spaces or line breaks counts as one space.

Source: FAQ: Indic Scripts and Languages, Unicode Consortium. Checked 26 October 2026.

Timing rules

• The clock starts with your first key, not when the page opens.
• If you finish the passage before time is up, your speed uses the time you actually took.
• If time runs out in the middle of a word, that unfinished word is left out of every count.
• Results need at least one second of typing; shorter times are treated as one second.

The extra views in your result

Word-count view

A CBSE document on its typing test formula counts whole words instead of 5-character words: gross words per minute is all typed words divided by minutes, net words per minute is correct words divided by minutes, and accuracy is NWPM × 100 ÷ GWPM. Your result shows these figures too. In the English example above: 8 typed words ÷ 0.25 = 32 GWPM, 7 correct words ÷ 0.25 = 28 NWPM, accuracy 87.5%.

Source: Typing Test Formula and Illustrations, Central Board of Secondary Education (CBSE). Checked 26 October 2026.

Characters per hour

SSC states data entry standards in key depressions per hour. Your result shows gross and net characters per hour (characters ÷ minutes × 60) for a rough comparison. They are not the same thing as key depressions: a capital letter or a Hindi letter on the Shift layer is one character but two key presses, and the official count is made by SSC on its own passage.

Mistake tally

SSC's evaluation guidelines group typing errors into full mistakes (such as omitted, added or misspelt words) and half mistakes (such as capitalisation and punctuation slips). Your result groups your mistakes in the same way as an approximation. It cannot spot spacing errors, swapped words or paragraph errors: a missing space (Ihope) shows up as one wrong word plus one missed word. TypeTez does not calculate an official error percentage. What the SSC documents say.

Source: Revised guidelines for evaluation of Typing Test/DEST scripts (English/Hindi), Staff Selection Commission, Eastern Region (Kolkata). Checked 26 October 2026.

What TypeTez does not do

• It does not reproduce any exam's software, passages, marking or cut-offs, and it is not an official simulator.
• It counts characters, not physical key presses, and cannot see which keyboard layout you used.
• It treats capitalisation and punctuation slips as wrong words for net speed, which is stricter than a half-mistake rule.